Introduction
Updated: February 20 , 2008
This page is maintained as a reverse-chronological experience report (starting in the Mini Reviews section below).
Buy links
Please see the Recommended page.
Testing hard disk speed and reliability
Sooner or later every digital photographer either fills up the hard drive and/or finds that performance is just becoming too slow. That’s when it’s time to consider a higher-performance solution, such as a striped RAID system. If you’re in the market for such a system, take a look at the DiskTester page which lists a number of recent reviews on this topic, all of which utilize DiskTester.
DiskTester also has a feature which can test for hard
drive reliability, in case you suspect
sporadic errors. It’s a wise precaution to run an overnight reliability test (or even 24 hours) before trusting
a new setup with your data.
Favorite hard drives
2008 favorite: Seagate ES.2 ST31000340NS
The Seagate 1TB (terabyte 1000MB) drive is clearly the best hard drive I’ve ever used. Quiet and fast, it can’t be beat. If you’re pressed for cash there is the ST31000340AS model, but if you actually need a 1TB hard drive you probably care about your data and you might as well pay the small premium for the ES.2 enterprise model (ST31000340NS), especially if you’re building a RAID.
See the DiskTester results.
2007 favorite: Maxtor Maxline III 7V300F0 300GB hard drive
[Spring/Summer/Fall 2006, see also Maxline Pro 7H500F0 below]
The Maxtor Maxline
III 7V300F0 [buy
See also AMUG’s report on the 7V300F0 (scroll to the middle of the article). AMUG’s November 2006 report is on the 320GB Maxtor 7V320F0, which is marginally less expensive, but also a bit slower. Considering the overall cost for an external setup (enclosure, SATA card, cables, etc), saving $10 or $15 makes little sense. Stick with the tried-and-true.
The site barefeats.com is also an excellent source of performance data.
2006 favorite: Maxtor Maxline III 7V300F0 300GB hard drive
[Spring/Summer/Fall 2006, see also Maxline Pro 7H500F0 below]
The Maxtor Maxline
III 7V300F0 [buy
See also AMUG’s report on the 7V300F0 (scroll to the middle of the article). AMUG’s November 2006 report is on the 320GB Maxtor 7V320F0, which is marginally less expensive, but also a bit slower. Considering the overall cost for an external setup (enclosure, SATA card, cables, etc), saving $10 or $15 makes little sense. Stick with the tried-and-true.
The site barefeats.com is also an excellent source of performance data.
Favorite: Maxtor Maxline Pro 7H500F0 500GB hard drive
[December, 2006]
The Maxtor Maxline
III 7H500F0 [buy
Look for rebates—I purchased a number of them in mid-December 2006 at newegg.com for $129 + tax with free shipping (offer no longer available). This is a good reason to belong to AMUG (besides their reviews); they send out news of discount codes on a regular basis.
Mini Reviews
Feb 20, 2008—Seagate 1TB drives
If you’re using these drives externally, be sure to use the latest drivers (eg FirmTek or Sonnet et al). I did not do so initially, and saw a puzzling 7MB/sec write speed (yes, 7 megabytes per second). The updated drivers resolved the issue.
Test results are with the Seagate ES.2, model ST31000340NS, using diglloyd DiskTester 2.0 on Mac OS X 10.4.11. Note that the actual formatted size is the usual nonsense of only a percentage of the alleged drive size, 931GB in this case.
SINGLE DRIVE, FirmTek SeriTek 2SE2-E SATA card
Let’s first look at single-drive speed using the FirmTek SeriTek 2SE2-E SATA card. Detailed results are shown below, but can be summarized as simply the best ever seen by this user. A year ago 80MB/sec would have been considered excellent, so 106MB/sec is really outstanding. All hard drives decline in speed across the drive (the simple math of data density X circumference).
disktester run-area-test -c 32M -t 8G -i 3 TB1 DiskTester 2.0 (C) 2003-2006 diglloyd, Inc. All Rights Reserve ... progress omitted ... ---------------- Averages for "TB1" (8GB/32MB, 3 iterations) --------------- Area (926.1GB) Write MB/sec Read MB/sec 0% 105.9 106.7 10% 103.8 104.1 20% 102.3 102.8 30% 98.2 98.9 40% 91.5 92.4 50% 88.1 89.0 60% 78.8 79.5 70% 75.1 75.8 80% 68.7 69.3 90% 59.9 60.5 100% 51.8 51.8 Average write speed across the volume: 84.0MB/sec Average read speed across the volume: 84.6MB/sec
4-DRIVE striped RAID
4-way striped RAID for “master” data drive
Remainder of space is used for two 2-way stripes “Back1” and “Back2”
(click to see larger)
If you use two of the FirmTek SeriTek 2SE2-E SATA cards for a 4-drive striped RAID, you’ll be disappointed to see that the cards (not the drives) are the limiting factor, but see below for the Sonnet Temp E4P results.
disktester run-area-test -c 32M -t 8G -i 5 TBx4 ... progress omitted ... ----------------- Averages for "TBx4" (8GB/32MB, 3 iterations) (FirmTek) Area (3.63TB) Write MB/sec Read MB/sec 0% 290 286 <====== limited by PCI card speed 10% 311 286 20% 314 286 30% 312 285 40% 313 287 50% 311 286 60% 307 286 70% 294 285 80% 267 269 90% 234 237 100% 204 203 Average write speed across the volume: 287MB/sec Average read speed across the volume: 272MB/sec
However, using the Sonnet Technologies Tempo E4P card the full potential is reached, showing linear scalability over a single drive. Even on the very slowest portion of the volume, 200MB/sec is maintained!
disktester run-area-test -c 32M -t 8G -i 5 TBx4
----------------- Averages for "TBx4" (8GB/32MB, 5 iterations) (Sonnet)
Area (3.63TB) Write MB/sec Read MB/sec
0% 415 420
10% 419 421
20% 402 406
30% 380 384
40% 351 356
50% 341 345
60% 310 313
70% 294 297
80% 266 269
90% 233 237
100% 202 203
Average write speed across the volume: 329MB/sec
Average read speed across the volume: 332MB/sec
Nov 29, 2006—SATA vs FireWire 800 on MacBook Pro Core 2 Duo
FirmTek sent me the driver software via email for their SeriTek/2SM2-E card. Inexplicably, it is not available on their web site! Without the driver software, the 2EN2 card does nothing at all.
How does FireWire 800 compare to SATA using the same drive (Maxtor 7V300F0) in the same enclosure (Wiebetech ToughTech XE )? Very favorably, it seems.
Firewire 800 is slightly slower for writes on the first 30% of the drive. Reads are just barely slower in that first 30%. After that (30% to 70%), speeds are identical (within the limits of normal variation).
SATA
vs FireWire 800 on MacBook Pro Core 2 Duo
[DiskTester data: SATA, Firewire
800]
click for larger image
For those looking to squeeze out every drop performance, using the SATA interface might be worthwhile, espeically if a partition is created out of the first portion of the drive (something less than the first 30%).
If a 2-drive system is used, then the SATA interface will be at least 50% faster, since FireWire 800 has a maximum theoretical speed of 100MB/sec., whereas dual drives (SeriTek/2EN2) through an SATA ExpressCard (SeriTek/2SM2-E) can achieve real speed of about 150MB/sec (see Nov 23 test).
Nov 26, 2006—FireWire 800 on MacBook Pro Core 2 Duo
Apple has long had a defective implementation of FireWire 800, with greatly impaired write speeds (see Nov 23 test below). While the PowerMac G4 series offered excellent performance (within the limits of FireWire 800), the PowerMac G5 series introduced very poor write speeds for both built-in and PCI-X cards with FireWire.
Now it looks like Apple has finally fixed FireWire 800 on the MacBook Pro Core 2 Duo. I used the same Wiebetech case with the Maxtor 7V300F0 drive installed as in the Nov 23rd test. Performance is excellent with FireWire 800 (I intended to also test with SATA, but the FirmTek SeriTek/2SM2-E card was not working with the MBP C2D, due to lack of driver software—see Nov 29 test for an update).
Note how FireWire 400 performance on the 2.16 GHz MacBook Pro can’t even reach the level of the worst performance of FireWire 800 on the 2.33 GHz MacBook Pro Core 2 Duo.
click for larger image
The 17" model of the original MacBook Pro offered a FireWire 800 port, but my 15" model doesn’t have one, so I could not verify if FireWire 800 is fixed on the original MacBook Pro.
MacBook Pro Core 2 Duo DiskTester Results—FireWire 800 [full
results]
“disktester run-area-test --iterations 3 --chunk-size 32M --test-size 2G BackupDuo”
------------- Averages for "Maxtor7V300F0" (2GB/32MB, 3 iterations) ------------
Area (277.4GB) Write MB/sec Read MB/sec
0% 66.0 70.2
10% 64.9 69.1
20% 65.6 66.9
30% 64.9 65.1
40% 64.3 63.2
50% 60.3 59.2
60% 58.4 57.2
70% 53.3 52.3
80% 50.8 49.9
90% 45.9 45.1
100% 39.8 39.2
Average write speed across the volume: 57.7MB/sec
Average read speed across the volume: 58.0MB/sec
Command "run-area-test" executed in 2411.92 seconds.
Nov 23, 2006—Maxtor 7V300F0 in Wiebetech case
SATA vs FireWire 800 vs USB 2.0
How does SATA compare to Firewire 800 and USB 2 for hard drive performance? Wiebetech’s ToughTech XE offers all 3 types of ports, and accepts SATA drives internally.
A Maxtor 7V300F0, one of the fastest SATA hard drives available (and with a 16MB onboard cache), was installed in the Toughtech XE case. DiskTester was used to measure sustained throughput with all three interfaces. (DiskTester is used by barefeats.com, amug.org and several vendors of SATA cards).
The results are unequivocal: on a PowerMac G5, stick with SATA unless you need the flexibility of FireWire. Avoid USB 2.0, unless there is no other option!
Click
for larger image
PowerMac Quad G5 DiskTester Results—SATA [full
results]
“disktester run-area-test --iterations 3 --chunk-size 32M --test-size 2G BackupDuo”
---------------- Averages for "Single" (2GB/32MB, 3 iterations) ----------------
Area (277.4GB) Write MB/sec Read MB/sec
0% 70.4 70.0
10% 69.4 69.0
20% 67.6 66.8
30% 65.8 65.0
40% 64.2 63.4
50% 60.0 59.2
60% 58.2 57.4
70% 53.1 52.5
80% 50.6 50.2
90% 45.7 45.4
100% 39.5 39.3
Average write speed across the volume: 58.6MB/sec
Average read speed across the volume: 58.0MB/sec
Command "run-area-test" executed in 2406.59 seconds
Apple has for the last several years (beginning with the G5 models) incompetently implemented Firewire 800 support. Performance is greatly degraded over the previous G4 models. The main symptom is greatly impaired write speed, approximately 42% of what it ought to be, but even read speed is impaired. The results below are not peculiar to the Maxtor 7V300F0 or the ToughTech XE case; poor performance is an invariant, even with 3rd-party FireWire cards.
PowerMac Quad G5 DiskTester Results—FireWire
800 [full results]
“disktester run-area-test --iterations 3 --chunk-size 32M --test-size 2G BackupDuo”
---------------- Averages for "Single" (2GB/32MB, 3 iterations) ----------------
Area (277.4GB) Write MB/sec Read MB/sec
0% 29.7 56.5
10% 29.7 56.6
20% 29.7 56.6
30% 29.5 56.1
40% 29.5 56.2
50% 29.5 56.2
60% 27.8 55.1
70% 29.3 52.6
80% 29.3 50.2
90% 29.3 45.5
100% 29.3 39.4
Average write speed across the volume: 29.3MB/sec
Average read speed across the volume: 52.8MB/sec
Command "run-area-test" executed in 3606.75 seconds.
USB 2.0 performance is really quite awful, so poor that it’s 1/4 to 1/3 of the minimum drive speed. Compare the 205 minutes it took to complete the test vs 40 minutes for SATA (5X faster) and 60 minutes for FireWire 800 (3.4X faster). There’s little more to say—if you like sluggish performance, use USB 2.0. As with FireWire 800, such performance is invariant; test results are consistent with other similar tests performed by the author in the past.
PowerMac Quad G5 DiskTester Results—USB 2.0 [full
results]
“disktester run-area-test --iterations 3 --chunk-size 32M --test-size 2G BackupDuo”
==== Thursday, November 23, 2006 5:06:31 PM PST, volume "Single" (279.2GB) ===
---------------- Averages for "Single" (2GB/32MB, 3 iterations) ----------------
Area (277.4GB) Write MB/sec Read MB/sec
0% 10.5 11.4
10% 10.5 11.4
20% 10.5 11.4
30% 10.8 11.4
40% 10.9 11.4
50% 10.9 11.5
60% 10.8 11.4
70% 10.5 11.4
80% 10.5 11.4
90% 10.5 11.4
100% 10.5 11.4
Average write speed across the volume: 10.6MB/sec
Average read speed across the volume: 11.4MB/sec
Command "run-area-test" executed in 12320.98 seconds.
November 23, 2006—Maxtor MaxLine Pro 500GB 2-way stripe
My storage needs hit 500GB, which meant that one of my backup systems was due for a capacity upgrade. I purchased two Maxtor MaxLine Pro 7H500F0 500GB bare drives at zipzoomfly.com ($239.99 each), and installed them into a FirmTek SeriTek/2EN2 (an exceptionally well-built 2-drive external SATA enclosure).
A striped RAID volume was created with the two new drives, using Mac OS X Disk Utility:
2-way striped RAID volume
Reliability
It is always wise to run the DiskTester test-reliability command (“disktester test-reliability BackupDuo”) overnight on new volumes that are to be put into production No errors were detected over an 8-hour continuous test.
Performance
The new 2-way striped RAID volume (“BackupDuo”) was tested for performance on the Macbook Pro using the FirmTek SeriTek/2SM2-E ExpressCard 34 adapter, which provides two external SATA II ports. An identical test was run on the PowerMac G5 Quad (using the Sonnet Tempo E4P PCI-X card).
The results are somewhat disappointing, being about 25% slower for writes than the 300GB Maxtor 7v300F0 (see Feb 16 Maxtor 7V300F0 results ). Though a different enclosure was used, and Max OS X 10.4.8 vs Mac OS X 10.4.2, the slower results are not likely to be due to those factors.
In the figures below, observe how performance drops on the outer tracks of the drives; speed is only about 60% of maximum as data is read from the outer tracks:
Performance declines across the volume whether you use a single drive or a striped RAID (though striped RAIDs that are limited by the bandwidth of the connection might show constant speed for most or all of the volume eg a 8-12 disk striped RAID).
With a 2-drive striped RAID, worst-case performance still remains well above that of the maximum performance of a single drive. When backing up 500GB of data, a 4-way stripe can accomplish the task in about 90 minutes; on a single drive about 6 hours is required —long enough to discourage regular backups.
MacBook Pro DiskTester Results
“disktester run-area-test --iterations 3 --chunk-size 32M --test-size 2G BackupDuo”
---------- Averages for "BackupDuo" (2GB/32MB, 3 iterations) ---------
Area (928.8GB) Write MB/sec Read MB/sec
0% 122.3 132.4
10% 127.3 131.9
20% 128.3 131.0
30% 126.4 126.8
40% 125.2 124.1
50% 122.0 120.9
60% 114.4 113.4
70% 109.2 107.2
80% 100.1 98.5
90% 89.2 87.8
100% 75.2 74.6
Average write speed across the volume: 113MB/sec
Average read speed across the volume: 114MB/sec
Command "run-area-test" executed in 1238.39 seconds.
PowerMac G5 Quad DiskTester Results
“disktester run-area-test --iterations 3 --chunk-size 32M --test-size 2G BackupDuo”
--------------- Averages for "BackupDuo" (2GB/32MB, 3 iterations) --------------
Area (928.8GB) Write MB/sec Read MB/sec
0% 122.3 133.9
10% 126.5 133.7
20% 128.7 131.1
30% 126.7 126.6
40% 125.6 124.5
50% 121.9 121.7
60% 114.0 114.0
70% 108.9 108.0
80% 99.7 99.0
90% 89.0 88.3
100% 75.2 74.9
Average write speed across the volume: 113MB/sec
Average read speed across the volume: 114MB/sec
Command "run-area-test" executed in 1236.13 seconds.
Feb 16, 2006 Maxtor 7V300F0 300GB
The new Maxtor 7V300F0 300GB hard drive might well be the fastest SATA drive you can buy today (Feb 2006), or at least on par with competitors like the Western Digital Raptor. Using DiskTester 1.0, a striped pair of these drives was tested on the PowerMac Quad, using the built-in SATA connectors. The results shown below are exceptional.
Results from DiskTester version 1.0
Conclusions
See also
PowerMac G5 Internal Drive Kits
MacBook Pro Experience Report
Sonnet Tempo ”Interrupt Storm” problem
Links
barefeats.com [Laptop
hard-drive shootout, TransIntl Mini Express, numerous
others]
amug.org [ Maxtor
7V320F0,
Maxtor DiamondMax]
Contact: email a comment on this article.
Enjoyed this article? Please use an affiliate link to help support the free content on this site. Affiliates include amazon.com and the Apple Store. Also, the Recommended Products and Vendors page offers discounts to diglloyd.com readers on certain products.
You’ve just enjoyed a free digloyd.com article. Diglloyd.com also offers paid articles/reviews, whose depth and breadth you’ll rarely find anywhere else—and which you can rely on for purchasing decisions.
Save yourself the aggravation and expense of buying the wrong equipment, and make the right choice for your photographic needs—get the reviews that analyze, inform, describe, and teach based on real-world experience, while providing detailed examples for you to make your own informed decisions.
 |
 |
 |
 |
 |
| Nikon D200 vs D2X | Nikon D2X vs Canon EOS | Raw-file Converters | The Sharpest Image | 28mm Shift Lenses |
© Copyright 2006 digLloyd, Inc. All Rights Reserved. Reproduction in any form requires explicit, written permission. No part of this document may be copied, distributed, disseminated in any form, electronic or tangible, whether in whole or in part, without the prior written authorization of digLloyd, Inc or its authorized representative. [Last updated: 12 November 2006]